]>
Commit | Line | Data |
---|---|---|
6c510bee | 1 | #include "cache.h" |
b2141fc1 | 2 | #include "config.h" |
cbd53a21 | 3 | #include "object-store.h" |
35ebfd6a | 4 | #include "attr.h" |
3fed15f5 | 5 | #include "run-command.h" |
a2b665de | 6 | #include "quote.h" |
6424c2ad | 7 | #include "sigchain.h" |
edcc8581 | 8 | #include "pkt-line.h" |
99605d62 | 9 | #include "sub-process.h" |
107642fe | 10 | #include "utf8.h" |
2c65d90f | 11 | #include "ll-merge.h" |
35ebfd6a | 12 | |
6c510bee LT |
13 | /* |
14 | * convert.c - convert a file when checking it out and checking it in. | |
15 | * | |
16 | * This should use the pathname to decide on whether it wants to do some | |
17 | * more interesting conversions (automatic gzip/unzip, general format | |
18 | * conversions etc etc), but by default it just does automatic CRLF<->LF | |
942e7747 | 19 | * translation when the "text" attribute or "auto_crlf" option is set. |
6c510bee LT |
20 | */ |
21 | ||
a7630bd4 TB |
22 | /* Stat bits: When BIN is set, the txt bits are unset */ |
23 | #define CONVERT_STAT_BITS_TXT_LF 0x1 | |
24 | #define CONVERT_STAT_BITS_TXT_CRLF 0x2 | |
25 | #define CONVERT_STAT_BITS_BIN 0x4 | |
26 | ||
6c510bee | 27 | struct text_stat { |
28624193 | 28 | /* NUL, CR, LF and CRLF counts */ |
6e336a53 | 29 | unsigned nul, lonecr, lonelf, crlf; |
6c510bee LT |
30 | |
31 | /* These are just approximations! */ | |
32 | unsigned printable, nonprintable; | |
33 | }; | |
34 | ||
35 | static void gather_stats(const char *buf, unsigned long size, struct text_stat *stats) | |
36 | { | |
37 | unsigned long i; | |
38 | ||
39 | memset(stats, 0, sizeof(*stats)); | |
40 | ||
41 | for (i = 0; i < size; i++) { | |
42 | unsigned char c = buf[i]; | |
43 | if (c == '\r') { | |
6e336a53 | 44 | if (i+1 < size && buf[i+1] == '\n') { |
6c510bee | 45 | stats->crlf++; |
6e336a53 TB |
46 | i++; |
47 | } else | |
48 | stats->lonecr++; | |
6c510bee LT |
49 | continue; |
50 | } | |
51 | if (c == '\n') { | |
6e336a53 | 52 | stats->lonelf++; |
6c510bee LT |
53 | continue; |
54 | } | |
55 | if (c == 127) | |
56 | /* DEL */ | |
57 | stats->nonprintable++; | |
58 | else if (c < 32) { | |
59 | switch (c) { | |
60 | /* BS, HT, ESC and FF */ | |
61 | case '\b': case '\t': case '\033': case '\014': | |
62 | stats->printable++; | |
63 | break; | |
28624193 DP |
64 | case 0: |
65 | stats->nul++; | |
66 | /* fall through */ | |
6c510bee LT |
67 | default: |
68 | stats->nonprintable++; | |
69 | } | |
70 | } | |
71 | else | |
72 | stats->printable++; | |
73 | } | |
f9dd4bf4 DK |
74 | |
75 | /* If file ends with EOF then don't count this EOF as non-printable. */ | |
76 | if (size >= 1 && buf[size-1] == '\032') | |
77 | stats->nonprintable--; | |
6c510bee LT |
78 | } |
79 | ||
80 | /* | |
81 | * The same heuristics as diff.c::mmfile_is_binary() | |
a7630bd4 | 82 | * We treat files with bare CR as binary |
6c510bee | 83 | */ |
129beeee | 84 | static int convert_is_binary(const struct text_stat *stats) |
6c510bee | 85 | { |
6e336a53 | 86 | if (stats->lonecr) |
a7630bd4 | 87 | return 1; |
28624193 DP |
88 | if (stats->nul) |
89 | return 1; | |
6c510bee LT |
90 | if ((stats->printable >> 7) < stats->nonprintable) |
91 | return 1; | |
6c510bee LT |
92 | return 0; |
93 | } | |
94 | ||
a7630bd4 TB |
95 | static unsigned int gather_convert_stats(const char *data, unsigned long size) |
96 | { | |
97 | struct text_stat stats; | |
6e336a53 | 98 | int ret = 0; |
a7630bd4 TB |
99 | if (!data || !size) |
100 | return 0; | |
101 | gather_stats(data, size, &stats); | |
129beeee | 102 | if (convert_is_binary(&stats)) |
6e336a53 TB |
103 | ret |= CONVERT_STAT_BITS_BIN; |
104 | if (stats.crlf) | |
105 | ret |= CONVERT_STAT_BITS_TXT_CRLF; | |
106 | if (stats.lonelf) | |
107 | ret |= CONVERT_STAT_BITS_TXT_LF; | |
108 | ||
109 | return ret; | |
a7630bd4 TB |
110 | } |
111 | ||
112 | static const char *gather_convert_stats_ascii(const char *data, unsigned long size) | |
113 | { | |
114 | unsigned int convert_stats = gather_convert_stats(data, size); | |
115 | ||
116 | if (convert_stats & CONVERT_STAT_BITS_BIN) | |
117 | return "-text"; | |
118 | switch (convert_stats) { | |
119 | case CONVERT_STAT_BITS_TXT_LF: | |
120 | return "lf"; | |
121 | case CONVERT_STAT_BITS_TXT_CRLF: | |
122 | return "crlf"; | |
123 | case CONVERT_STAT_BITS_TXT_LF | CONVERT_STAT_BITS_TXT_CRLF: | |
124 | return "mixed"; | |
125 | default: | |
126 | return "none"; | |
127 | } | |
128 | } | |
129 | ||
847a9e5d | 130 | const char *get_cached_convert_stats_ascii(struct index_state *istate, |
a7609c54 | 131 | const char *path) |
a7630bd4 TB |
132 | { |
133 | const char *ret; | |
134 | unsigned long sz; | |
a7609c54 | 135 | void *data = read_blob_data_from_index(istate, path, &sz); |
a7630bd4 TB |
136 | ret = gather_convert_stats_ascii(data, sz); |
137 | free(data); | |
138 | return ret; | |
139 | } | |
140 | ||
141 | const char *get_wt_convert_stats_ascii(const char *path) | |
142 | { | |
143 | const char *ret = ""; | |
144 | struct strbuf sb = STRBUF_INIT; | |
145 | if (strbuf_read_file(&sb, path, 0) >= 0) | |
146 | ret = gather_convert_stats_ascii(sb.buf, sb.len); | |
147 | strbuf_release(&sb); | |
148 | return ret; | |
149 | } | |
150 | ||
4b4024f5 TB |
151 | static int text_eol_is_crlf(void) |
152 | { | |
153 | if (auto_crlf == AUTO_CRLF_TRUE) | |
154 | return 1; | |
155 | else if (auto_crlf == AUTO_CRLF_INPUT) | |
156 | return 0; | |
157 | if (core_eol == EOL_CRLF) | |
158 | return 1; | |
159 | if (core_eol == EOL_UNSET && EOL_NATIVE == EOL_CRLF) | |
160 | return 1; | |
161 | return 0; | |
162 | } | |
163 | ||
38e95844 | 164 | static enum eol output_eol(enum convert_crlf_action crlf_action) |
f217f0e8 | 165 | { |
c61dcff9 | 166 | switch (crlf_action) { |
942e7747 EB |
167 | case CRLF_BINARY: |
168 | return EOL_UNSET; | |
df747b81 | 169 | case CRLF_TEXT_CRLF: |
942e7747 | 170 | return EOL_CRLF; |
df747b81 | 171 | case CRLF_TEXT_INPUT: |
942e7747 | 172 | return EOL_LF; |
df747b81 TB |
173 | case CRLF_UNDEFINED: |
174 | case CRLF_AUTO_CRLF: | |
65237284 | 175 | return EOL_CRLF; |
df747b81 | 176 | case CRLF_AUTO_INPUT: |
65237284 | 177 | return EOL_LF; |
942e7747 EB |
178 | case CRLF_TEXT: |
179 | case CRLF_AUTO: | |
df747b81 | 180 | /* fall through */ |
4b4024f5 | 181 | return text_eol_is_crlf() ? EOL_CRLF : EOL_LF; |
942e7747 | 182 | } |
d26a328e | 183 | warning(_("illegal crlf_action %d"), (int)crlf_action); |
ec70f52f | 184 | return core_eol; |
942e7747 EB |
185 | } |
186 | ||
185e8652 | 187 | static void check_global_conv_flags_eol(const char *path, |
a0ad53c1 | 188 | struct text_stat *old_stats, struct text_stat *new_stats, |
8462ff43 | 189 | int conv_flags) |
21e5ad50 | 190 | { |
a0ad53c1 | 191 | if (old_stats->crlf && !new_stats->crlf ) { |
21e5ad50 | 192 | /* |
a0ad53c1 | 193 | * CRLFs would not be restored by checkout |
21e5ad50 | 194 | */ |
8462ff43 | 195 | if (conv_flags & CONV_EOL_RNDTRP_DIE) |
1a07e59c | 196 | die(_("CRLF would be replaced by LF in %s"), path); |
8462ff43 | 197 | else if (conv_flags & CONV_EOL_RNDTRP_WARN) |
87cb7845 VA |
198 | warning(_("CRLF will be replaced by LF in %s.\n" |
199 | "The file will have its original line" | |
d26a328e | 200 | " endings in your working directory"), path); |
a0ad53c1 | 201 | } else if (old_stats->lonelf && !new_stats->lonelf ) { |
21e5ad50 | 202 | /* |
a0ad53c1 | 203 | * CRLFs would be added by checkout |
21e5ad50 | 204 | */ |
8462ff43 TB |
205 | if (conv_flags & CONV_EOL_RNDTRP_DIE) |
206 | die(_("LF would be replaced by CRLF in %s"), path); | |
207 | else if (conv_flags & CONV_EOL_RNDTRP_WARN) | |
87cb7845 VA |
208 | warning(_("LF will be replaced by CRLF in %s.\n" |
209 | "The file will have its original line" | |
d26a328e | 210 | " endings in your working directory"), path); |
21e5ad50 SP |
211 | } |
212 | } | |
213 | ||
847a9e5d | 214 | static int has_crlf_in_index(struct index_state *istate, const char *path) |
c4805393 | 215 | { |
c4805393 | 216 | unsigned long sz; |
c4805393 | 217 | void *data; |
86ff70a0 TB |
218 | const char *crp; |
219 | int has_crlf = 0; | |
c4805393 | 220 | |
49a6d31f | 221 | data = read_blob_data_from_index(istate, path, &sz); |
4982fd78 | 222 | if (!data) |
c4805393 | 223 | return 0; |
86ff70a0 TB |
224 | |
225 | crp = memchr(data, '\r', sz); | |
226 | if (crp) { | |
227 | unsigned int ret_stats; | |
228 | ret_stats = gather_convert_stats(data, sz); | |
229 | if (!(ret_stats & CONVERT_STAT_BITS_BIN) && | |
230 | (ret_stats & CONVERT_STAT_BITS_TXT_CRLF)) | |
231 | has_crlf = 1; | |
232 | } | |
c4805393 | 233 | free(data); |
86ff70a0 | 234 | return has_crlf; |
c4805393 FAG |
235 | } |
236 | ||
129beeee | 237 | static int will_convert_lf_to_crlf(struct text_stat *stats, |
38e95844 | 238 | enum convert_crlf_action crlf_action) |
a0ad53c1 TB |
239 | { |
240 | if (output_eol(crlf_action) != EOL_CRLF) | |
241 | return 0; | |
242 | /* No "naked" LF? Nothing to convert, regardless. */ | |
243 | if (!stats->lonelf) | |
244 | return 0; | |
245 | ||
246 | if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) { | |
247 | /* If we have any CR or CRLF line endings, we do not touch it */ | |
248 | /* This is the new safer autocrlf-handling */ | |
249 | if (stats->lonecr || stats->crlf) | |
250 | return 0; | |
251 | ||
129beeee | 252 | if (convert_is_binary(stats)) |
a0ad53c1 TB |
253 | return 0; |
254 | } | |
255 | return 1; | |
256 | ||
257 | } | |
258 | ||
7a17918c LS |
259 | static int validate_encoding(const char *path, const char *enc, |
260 | const char *data, size_t len, int die_on_error) | |
261 | { | |
ed283588 RS |
262 | const char *stripped; |
263 | ||
7a17918c | 264 | /* We only check for UTF here as UTF?? can be an alias for UTF-?? */ |
ed283588 RS |
265 | if (skip_iprefix(enc, "UTF", &stripped)) { |
266 | skip_prefix(stripped, "-", &stripped); | |
267 | ||
7a17918c LS |
268 | /* |
269 | * Check for detectable errors in UTF encodings | |
270 | */ | |
271 | if (has_prohibited_utf_bom(enc, data, len)) { | |
272 | const char *error_msg = _( | |
273 | "BOM is prohibited in '%s' if encoded as %s"); | |
274 | /* | |
275 | * This advice is shown for UTF-??BE and UTF-??LE encodings. | |
276 | * We cut off the last two characters of the encoding name | |
277 | * to generate the encoding name suitable for BOMs. | |
278 | */ | |
279 | const char *advise_msg = _( | |
280 | "The file '%s' contains a byte order " | |
ed283588 | 281 | "mark (BOM). Please use UTF-%.*s as " |
7a17918c | 282 | "working-tree-encoding."); |
ed283588 RS |
283 | int stripped_len = strlen(stripped) - strlen("BE"); |
284 | advise(advise_msg, path, stripped_len, stripped); | |
7a17918c LS |
285 | if (die_on_error) |
286 | die(error_msg, path, enc); | |
287 | else { | |
288 | return error(error_msg, path, enc); | |
289 | } | |
290 | ||
291 | } else if (is_missing_required_utf_bom(enc, data, len)) { | |
292 | const char *error_msg = _( | |
293 | "BOM is required in '%s' if encoded as %s"); | |
294 | const char *advise_msg = _( | |
295 | "The file '%s' is missing a byte order " | |
296 | "mark (BOM). Please use UTF-%sBE or UTF-%sLE " | |
297 | "(depending on the byte order) as " | |
298 | "working-tree-encoding."); | |
7a17918c | 299 | advise(advise_msg, path, stripped, stripped); |
7a17918c LS |
300 | if (die_on_error) |
301 | die(error_msg, path, enc); | |
302 | else { | |
303 | return error(error_msg, path, enc); | |
304 | } | |
305 | } | |
306 | ||
307 | } | |
308 | return 0; | |
309 | } | |
310 | ||
541d059c LS |
311 | static void trace_encoding(const char *context, const char *path, |
312 | const char *encoding, const char *buf, size_t len) | |
313 | { | |
314 | static struct trace_key coe = TRACE_KEY_INIT(WORKING_TREE_ENCODING); | |
315 | struct strbuf trace = STRBUF_INIT; | |
316 | int i; | |
317 | ||
318 | strbuf_addf(&trace, "%s (%s, considered %s):\n", context, path, encoding); | |
319 | for (i = 0; i < len && buf; ++i) { | |
320 | strbuf_addf( | |
8302f50e | 321 | &trace, "| \033[2m%2i:\033[0m %2x \033[2m%c\033[0m%c", |
541d059c LS |
322 | i, |
323 | (unsigned char) buf[i], | |
324 | (buf[i] > 32 && buf[i] < 127 ? buf[i] : ' '), | |
325 | ((i+1) % 8 && (i+1) < len ? ' ' : '\n') | |
326 | ); | |
327 | } | |
328 | strbuf_addchars(&trace, '\n', 1); | |
329 | ||
330 | trace_strbuf(&coe, &trace); | |
331 | strbuf_release(&trace); | |
332 | } | |
333 | ||
e92d6225 LS |
334 | static int check_roundtrip(const char *enc_name) |
335 | { | |
336 | /* | |
337 | * check_roundtrip_encoding contains a string of comma and/or | |
338 | * space separated encodings (eg. "UTF-16, ASCII, CP1125"). | |
339 | * Search for the given encoding in that string. | |
340 | */ | |
341 | const char *found = strcasestr(check_roundtrip_encoding, enc_name); | |
342 | const char *next; | |
343 | int len; | |
344 | if (!found) | |
345 | return 0; | |
346 | next = found + strlen(enc_name); | |
347 | len = strlen(check_roundtrip_encoding); | |
348 | return (found && ( | |
349 | /* | |
350 | * check that the found encoding is at the | |
351 | * beginning of check_roundtrip_encoding or | |
352 | * that it is prefixed with a space or comma | |
353 | */ | |
354 | found == check_roundtrip_encoding || ( | |
355 | (isspace(found[-1]) || found[-1] == ',') | |
356 | ) | |
357 | ) && ( | |
358 | /* | |
359 | * check that the found encoding is at the | |
360 | * end of check_roundtrip_encoding or | |
361 | * that it is suffixed with a space or comma | |
362 | */ | |
363 | next == check_roundtrip_encoding + len || ( | |
364 | next < check_roundtrip_encoding + len && | |
365 | (isspace(next[0]) || next[0] == ',') | |
366 | ) | |
367 | )); | |
368 | } | |
369 | ||
107642fe LS |
370 | static const char *default_encoding = "UTF-8"; |
371 | ||
372 | static int encode_to_git(const char *path, const char *src, size_t src_len, | |
373 | struct strbuf *buf, const char *enc, int conv_flags) | |
374 | { | |
375 | char *dst; | |
c7d017d7 | 376 | size_t dst_len; |
107642fe LS |
377 | int die_on_error = conv_flags & CONV_WRITE_OBJECT; |
378 | ||
379 | /* | |
380 | * No encoding is specified or there is nothing to encode. | |
381 | * Tell the caller that the content was not modified. | |
382 | */ | |
383 | if (!enc || (src && !src_len)) | |
384 | return 0; | |
385 | ||
386 | /* | |
387 | * Looks like we got called from "would_convert_to_git()". | |
388 | * This means Git wants to know if it would encode (= modify!) | |
389 | * the content. Let's answer with "yes", since an encoding was | |
390 | * specified. | |
391 | */ | |
392 | if (!buf && !src) | |
393 | return 1; | |
394 | ||
7a17918c LS |
395 | if (validate_encoding(path, enc, src, src_len, die_on_error)) |
396 | return 0; | |
397 | ||
541d059c | 398 | trace_encoding("source", path, enc, src, src_len); |
107642fe LS |
399 | dst = reencode_string_len(src, src_len, default_encoding, enc, |
400 | &dst_len); | |
401 | if (!dst) { | |
402 | /* | |
403 | * We could add the blob "as-is" to Git. However, on checkout | |
15beaaa3 | 404 | * we would try to re-encode to the original encoding. This |
107642fe LS |
405 | * would fail and we would leave the user with a messed-up |
406 | * working tree. Let's try to avoid this by screaming loud. | |
407 | */ | |
408 | const char* msg = _("failed to encode '%s' from %s to %s"); | |
409 | if (die_on_error) | |
410 | die(msg, path, enc, default_encoding); | |
411 | else { | |
412 | error(msg, path, enc, default_encoding); | |
413 | return 0; | |
414 | } | |
415 | } | |
541d059c | 416 | trace_encoding("destination", path, default_encoding, dst, dst_len); |
107642fe | 417 | |
e92d6225 LS |
418 | /* |
419 | * UTF supports lossless conversion round tripping [1] and conversions | |
420 | * between UTF and other encodings are mostly round trip safe as | |
421 | * Unicode aims to be a superset of all other character encodings. | |
422 | * However, certain encodings (e.g. SHIFT-JIS) are known to have round | |
423 | * trip issues [2]. Check the round trip conversion for all encodings | |
424 | * listed in core.checkRoundtripEncoding. | |
425 | * | |
426 | * The round trip check is only performed if content is written to Git. | |
427 | * This ensures that no information is lost during conversion to/from | |
428 | * the internal UTF-8 representation. | |
429 | * | |
430 | * Please note, the code below is not tested because I was not able to | |
431 | * generate a faulty round trip without an iconv error. Iconv errors | |
432 | * are already caught above. | |
433 | * | |
434 | * [1] http://unicode.org/faq/utf_bom.html#gen2 | |
435 | * [2] https://support.microsoft.com/en-us/help/170559/prb-conversion-problem-between-shift-jis-and-unicode | |
436 | */ | |
437 | if (die_on_error && check_roundtrip(enc)) { | |
438 | char *re_src; | |
c7d017d7 | 439 | size_t re_src_len; |
e92d6225 LS |
440 | |
441 | re_src = reencode_string_len(dst, dst_len, | |
442 | enc, default_encoding, | |
443 | &re_src_len); | |
444 | ||
445 | trace_printf("Checking roundtrip encoding for %s...\n", enc); | |
446 | trace_encoding("reencoded source", path, enc, | |
447 | re_src, re_src_len); | |
448 | ||
449 | if (!re_src || src_len != re_src_len || | |
450 | memcmp(src, re_src, src_len)) { | |
451 | const char* msg = _("encoding '%s' from %s to %s and " | |
452 | "back is not the same"); | |
453 | die(msg, path, enc, default_encoding); | |
454 | } | |
455 | ||
456 | free(re_src); | |
457 | } | |
458 | ||
107642fe LS |
459 | strbuf_attach(buf, dst, dst_len, dst_len + 1); |
460 | return 1; | |
461 | } | |
462 | ||
463 | static int encode_to_worktree(const char *path, const char *src, size_t src_len, | |
464 | struct strbuf *buf, const char *enc) | |
465 | { | |
466 | char *dst; | |
c7d017d7 | 467 | size_t dst_len; |
107642fe LS |
468 | |
469 | /* | |
470 | * No encoding is specified or there is nothing to encode. | |
471 | * Tell the caller that the content was not modified. | |
472 | */ | |
473 | if (!enc || (src && !src_len)) | |
474 | return 0; | |
475 | ||
476 | dst = reencode_string_len(src, src_len, enc, default_encoding, | |
477 | &dst_len); | |
478 | if (!dst) { | |
d26a328e | 479 | error(_("failed to encode '%s' from %s to %s"), |
1a07e59c | 480 | path, default_encoding, enc); |
107642fe LS |
481 | return 0; |
482 | } | |
483 | ||
484 | strbuf_attach(buf, dst, dst_len, dst_len + 1); | |
485 | return 1; | |
486 | } | |
487 | ||
847a9e5d | 488 | static int crlf_to_git(struct index_state *istate, |
49a6d31f | 489 | const char *path, const char *src, size_t len, |
c61dcff9 | 490 | struct strbuf *buf, |
38e95844 | 491 | enum convert_crlf_action crlf_action, int conv_flags) |
6c510bee | 492 | { |
6c510bee | 493 | struct text_stat stats; |
5ecd293d | 494 | char *dst; |
a0ad53c1 | 495 | int convert_crlf_into_lf; |
6c510bee | 496 | |
c61dcff9 | 497 | if (crlf_action == CRLF_BINARY || |
4c3b57b9 | 498 | (src && !len)) |
5ecd293d | 499 | return 0; |
6c510bee | 500 | |
4c3b57b9 JK |
501 | /* |
502 | * If we are doing a dry-run and have no source buffer, there is | |
503 | * nothing to analyze; we must assume we would convert. | |
504 | */ | |
505 | if (!buf && !src) | |
506 | return 1; | |
507 | ||
5ecd293d | 508 | gather_stats(src, len, &stats); |
a0ad53c1 TB |
509 | /* Optimization: No CRLF? Nothing to convert, regardless. */ |
510 | convert_crlf_into_lf = !!stats.crlf; | |
6c510bee | 511 | |
df747b81 | 512 | if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) { |
129beeee | 513 | if (convert_is_binary(&stats)) |
5ecd293d | 514 | return 0; |
65237284 | 515 | /* |
1c25d2d8 TB |
516 | * If the file in the index has any CR in it, do not |
517 | * convert. This is the new safer autocrlf handling, | |
518 | * unless we want to renormalize in a merge or | |
519 | * cherry-pick. | |
65237284 | 520 | */ |
8462ff43 | 521 | if ((!(conv_flags & CONV_EOL_RENORMALIZE)) && |
86ff70a0 | 522 | has_crlf_in_index(istate, path)) |
a0ad53c1 | 523 | convert_crlf_into_lf = 0; |
201ac8ef | 524 | } |
8462ff43 TB |
525 | if (((conv_flags & CONV_EOL_RNDTRP_WARN) || |
526 | ((conv_flags & CONV_EOL_RNDTRP_DIE) && len))) { | |
a0ad53c1 TB |
527 | struct text_stat new_stats; |
528 | memcpy(&new_stats, &stats, sizeof(new_stats)); | |
529 | /* simulate "git add" */ | |
530 | if (convert_crlf_into_lf) { | |
531 | new_stats.lonelf += new_stats.crlf; | |
532 | new_stats.crlf = 0; | |
533 | } | |
534 | /* simulate "git checkout" */ | |
129beeee | 535 | if (will_convert_lf_to_crlf(&new_stats, crlf_action)) { |
a0ad53c1 TB |
536 | new_stats.crlf += new_stats.lonelf; |
537 | new_stats.lonelf = 0; | |
538 | } | |
185e8652 | 539 | check_global_conv_flags_eol(path, &stats, &new_stats, conv_flags); |
a0ad53c1 TB |
540 | } |
541 | if (!convert_crlf_into_lf) | |
21e5ad50 SP |
542 | return 0; |
543 | ||
92ac3197 JK |
544 | /* |
545 | * At this point all of our source analysis is done, and we are sure we | |
546 | * would convert. If we are in dry-run mode, we can give an answer. | |
547 | */ | |
548 | if (!buf) | |
549 | return 1; | |
550 | ||
90d16ec0 PH |
551 | /* only grow if not in place */ |
552 | if (strbuf_avail(buf) + buf->len < len) | |
553 | strbuf_grow(buf, len - buf->len); | |
5ecd293d | 554 | dst = buf->buf; |
df747b81 | 555 | if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) { |
163b9591 JH |
556 | /* |
557 | * If we guessed, we already know we rejected a file with | |
558 | * lone CR, and we can strip a CR without looking at what | |
559 | * follow it. | |
560 | */ | |
201ac8ef | 561 | do { |
ac78e548 | 562 | unsigned char c = *src++; |
201ac8ef | 563 | if (c != '\r') |
ac78e548 | 564 | *dst++ = c; |
5ecd293d | 565 | } while (--len); |
201ac8ef JH |
566 | } else { |
567 | do { | |
ac78e548 | 568 | unsigned char c = *src++; |
5ecd293d | 569 | if (! (c == '\r' && (1 < len && *src == '\n'))) |
ac78e548 | 570 | *dst++ = c; |
5ecd293d | 571 | } while (--len); |
201ac8ef | 572 | } |
5ecd293d PH |
573 | strbuf_setlen(buf, dst - buf->buf); |
574 | return 1; | |
6c510bee LT |
575 | } |
576 | ||
38e95844 JH |
577 | static int crlf_to_worktree(const char *src, size_t len, struct strbuf *buf, |
578 | enum convert_crlf_action crlf_action) | |
6c510bee | 579 | { |
5ecd293d | 580 | char *to_free = NULL; |
6c510bee | 581 | struct text_stat stats; |
6c510bee | 582 | |
c61dcff9 | 583 | if (!len || output_eol(crlf_action) != EOL_CRLF) |
5ecd293d | 584 | return 0; |
6c510bee | 585 | |
5ecd293d | 586 | gather_stats(src, len, &stats); |
129beeee | 587 | if (!will_convert_lf_to_crlf(&stats, crlf_action)) |
5ecd293d | 588 | return 0; |
6c510bee | 589 | |
5ecd293d PH |
590 | /* are we "faking" in place editing ? */ |
591 | if (src == buf->buf) | |
b315c5c0 | 592 | to_free = strbuf_detach(buf, NULL); |
5ecd293d | 593 | |
6e336a53 | 594 | strbuf_grow(buf, len + stats.lonelf); |
5ecd293d PH |
595 | for (;;) { |
596 | const char *nl = memchr(src, '\n', len); | |
597 | if (!nl) | |
598 | break; | |
599 | if (nl > src && nl[-1] == '\r') { | |
600 | strbuf_add(buf, src, nl + 1 - src); | |
601 | } else { | |
602 | strbuf_add(buf, src, nl - src); | |
603 | strbuf_addstr(buf, "\r\n"); | |
604 | } | |
605 | len -= nl + 1 - src; | |
606 | src = nl + 1; | |
607 | } | |
608 | strbuf_add(buf, src, len); | |
609 | ||
610 | free(to_free); | |
611 | return 1; | |
6c510bee | 612 | } |
35ebfd6a | 613 | |
546bb582 JS |
614 | struct filter_params { |
615 | const char *src; | |
616 | unsigned long size; | |
9035d75a | 617 | int fd; |
546bb582 | 618 | const char *cmd; |
a2b665de | 619 | const char *path; |
546bb582 JS |
620 | }; |
621 | ||
9035d75a | 622 | static int filter_buffer_or_fd(int in, int out, void *data) |
aa4ed402 JH |
623 | { |
624 | /* | |
625 | * Spawn cmd and feed the buffer contents through its stdin. | |
626 | */ | |
d3180279 | 627 | struct child_process child_process = CHILD_PROCESS_INIT; |
546bb582 | 628 | struct filter_params *params = (struct filter_params *)data; |
aa4ed402 | 629 | int write_err, status; |
66dbfd55 | 630 | |
a2b665de PW |
631 | /* apply % substitution to cmd */ |
632 | struct strbuf cmd = STRBUF_INIT; | |
633 | struct strbuf path = STRBUF_INIT; | |
634 | struct strbuf_expand_dict_entry dict[] = { | |
635 | { "f", NULL, }, | |
636 | { NULL, NULL, }, | |
637 | }; | |
638 | ||
639 | /* quote the path to preserve spaces, etc. */ | |
640 | sq_quote_buf(&path, params->path); | |
641 | dict[0].value = path.buf; | |
642 | ||
643 | /* expand all %f with the quoted path */ | |
644 | strbuf_expand(&cmd, params->cmd, strbuf_expand_dict_cb, &dict); | |
645 | strbuf_release(&path); | |
646 | ||
afbdba39 | 647 | strvec_push(&child_process.args, cmd.buf); |
ac0ba18d | 648 | child_process.use_shell = 1; |
dc1bfdcd | 649 | child_process.in = -1; |
ae6a5609 | 650 | child_process.out = out; |
aa4ed402 | 651 | |
f31f1d39 RS |
652 | if (start_command(&child_process)) { |
653 | strbuf_release(&cmd); | |
d26a328e NTND |
654 | return error(_("cannot fork to run external filter '%s'"), |
655 | params->cmd); | |
f31f1d39 | 656 | } |
aa4ed402 | 657 | |
6424c2ad JB |
658 | sigchain_push(SIGPIPE, SIG_IGN); |
659 | ||
9035d75a | 660 | if (params->src) { |
0c4dd67a JH |
661 | write_err = (write_in_full(child_process.in, |
662 | params->src, params->size) < 0); | |
663 | if (errno == EPIPE) | |
664 | write_err = 0; | |
9035d75a SP |
665 | } else { |
666 | write_err = copy_fd(params->fd, child_process.in); | |
0c4dd67a JH |
667 | if (write_err == COPY_WRITE_ERROR && errno == EPIPE) |
668 | write_err = 0; | |
9035d75a SP |
669 | } |
670 | ||
dc1bfdcd | 671 | if (close(child_process.in)) |
aa4ed402 JH |
672 | write_err = 1; |
673 | if (write_err) | |
d26a328e NTND |
674 | error(_("cannot feed the input to external filter '%s'"), |
675 | params->cmd); | |
aa4ed402 | 676 | |
6424c2ad JB |
677 | sigchain_pop(SIGPIPE); |
678 | ||
aa4ed402 JH |
679 | status = finish_command(&child_process); |
680 | if (status) | |
d26a328e | 681 | error(_("external filter '%s' failed %d"), params->cmd, status); |
a2b665de PW |
682 | |
683 | strbuf_release(&cmd); | |
aa4ed402 JH |
684 | return (write_err || status); |
685 | } | |
686 | ||
234fa07e | 687 | static int apply_single_file_filter(const char *path, const char *src, size_t len, int fd, |
ec36c42a | 688 | struct strbuf *dst, const char *cmd) |
aa4ed402 JH |
689 | { |
690 | /* | |
691 | * Create a pipeline to have the command filter the buffer's | |
692 | * contents. | |
693 | * | |
694 | * (child --> cmd) --> us | |
695 | */ | |
b84be553 | 696 | int err = 0; |
f285a2d7 | 697 | struct strbuf nbuf = STRBUF_INIT; |
546bb582 JS |
698 | struct async async; |
699 | struct filter_params params; | |
aa4ed402 | 700 | |
546bb582 | 701 | memset(&async, 0, sizeof(async)); |
9035d75a | 702 | async.proc = filter_buffer_or_fd; |
546bb582 | 703 | async.data = ¶ms; |
ae6a5609 | 704 | async.out = -1; |
546bb582 JS |
705 | params.src = src; |
706 | params.size = len; | |
9035d75a | 707 | params.fd = fd; |
546bb582 | 708 | params.cmd = cmd; |
a2b665de | 709 | params.path = path; |
aa4ed402 JH |
710 | |
711 | fflush(NULL); | |
546bb582 JS |
712 | if (start_async(&async)) |
713 | return 0; /* error was already reported */ | |
aa4ed402 | 714 | |
02156ab0 | 715 | if (strbuf_read(&nbuf, async.out, 0) < 0) { |
d26a328e | 716 | err = error(_("read from external filter '%s' failed"), cmd); |
aa4ed402 | 717 | } |
546bb582 | 718 | if (close(async.out)) { |
d26a328e | 719 | err = error(_("read from external filter '%s' failed"), cmd); |
aa4ed402 | 720 | } |
546bb582 | 721 | if (finish_async(&async)) { |
d26a328e | 722 | err = error(_("external filter '%s' failed"), cmd); |
aa4ed402 JH |
723 | } |
724 | ||
b84be553 | 725 | if (!err) { |
90d16ec0 | 726 | strbuf_swap(dst, &nbuf); |
5ecd293d | 727 | } |
90d16ec0 | 728 | strbuf_release(&nbuf); |
b84be553 | 729 | return !err; |
aa4ed402 JH |
730 | } |
731 | ||
234fa07e LS |
732 | #define CAP_CLEAN (1u<<0) |
733 | #define CAP_SMUDGE (1u<<1) | |
2841e8f8 | 734 | #define CAP_DELAY (1u<<2) |
234fa07e | 735 | |
1b0b46ee BP |
736 | struct cmd2process { |
737 | struct subprocess_entry subprocess; /* must be the first member! */ | |
738 | unsigned int supported_capabilities; | |
739 | }; | |
740 | ||
f514d7d1 BP |
741 | static int subprocess_map_initialized; |
742 | static struct hashmap subprocess_map; | |
edcc8581 | 743 | |
7ddb9b2c | 744 | static int start_multi_file_filter_fn(struct subprocess_entry *subprocess) |
edcc8581 | 745 | { |
fa64a2fd JT |
746 | static int versions[] = {2, 0}; |
747 | static struct subprocess_capability capabilities[] = { | |
1514c8ed LS |
748 | { "clean", CAP_CLEAN }, |
749 | { "smudge", CAP_SMUDGE }, | |
2841e8f8 | 750 | { "delay", CAP_DELAY }, |
fa64a2fd | 751 | { NULL, 0 } |
1514c8ed | 752 | }; |
fa64a2fd JT |
753 | struct cmd2process *entry = (struct cmd2process *)subprocess; |
754 | return subprocess_handshake(subprocess, "git-filter", versions, NULL, | |
755 | capabilities, | |
756 | &entry->supported_capabilities); | |
a810ea99 BP |
757 | } |
758 | ||
9364fc29 LS |
759 | static void handle_filter_error(const struct strbuf *filter_status, |
760 | struct cmd2process *entry, | |
3b335762 NTND |
761 | const unsigned int wanted_capability) |
762 | { | |
9364fc29 LS |
763 | if (!strcmp(filter_status->buf, "error")) |
764 | ; /* The filter signaled a problem with the file. */ | |
765 | else if (!strcmp(filter_status->buf, "abort") && wanted_capability) { | |
766 | /* | |
767 | * The filter signaled a permanent problem. Don't try to filter | |
768 | * files with the same command for the lifetime of the current | |
769 | * Git process. | |
770 | */ | |
771 | entry->supported_capabilities &= ~wanted_capability; | |
772 | } else { | |
773 | /* | |
774 | * Something went wrong with the protocol filter. | |
775 | * Force shutdown and restart if another blob requires filtering. | |
776 | */ | |
d26a328e | 777 | error(_("external filter '%s' failed"), entry->subprocess.cmd); |
9364fc29 LS |
778 | subprocess_stop(&subprocess_map, &entry->subprocess); |
779 | free(entry); | |
780 | } | |
781 | } | |
782 | ||
edcc8581 LS |
783 | static int apply_multi_file_filter(const char *path, const char *src, size_t len, |
784 | int fd, struct strbuf *dst, const char *cmd, | |
2841e8f8 | 785 | const unsigned int wanted_capability, |
ab90ecae | 786 | const struct checkout_metadata *meta, |
2841e8f8 | 787 | struct delayed_checkout *dco) |
edcc8581 LS |
788 | { |
789 | int err; | |
2841e8f8 | 790 | int can_delay = 0; |
edcc8581 LS |
791 | struct cmd2process *entry; |
792 | struct child_process *process; | |
793 | struct strbuf nbuf = STRBUF_INIT; | |
794 | struct strbuf filter_status = STRBUF_INIT; | |
795 | const char *filter_type; | |
796 | ||
f514d7d1 BP |
797 | if (!subprocess_map_initialized) { |
798 | subprocess_map_initialized = 1; | |
9ab42958 | 799 | hashmap_init(&subprocess_map, cmd2process_cmp, NULL, 0); |
edcc8581 LS |
800 | entry = NULL; |
801 | } else { | |
f514d7d1 | 802 | entry = (struct cmd2process *)subprocess_find_entry(&subprocess_map, cmd); |
edcc8581 LS |
803 | } |
804 | ||
805 | fflush(NULL); | |
806 | ||
807 | if (!entry) { | |
7ddb9b2c BP |
808 | entry = xmalloc(sizeof(*entry)); |
809 | entry->supported_capabilities = 0; | |
810 | ||
f514d7d1 | 811 | if (subprocess_start(&subprocess_map, &entry->subprocess, cmd, start_multi_file_filter_fn)) { |
7ddb9b2c | 812 | free(entry); |
edcc8581 | 813 | return 0; |
7ddb9b2c | 814 | } |
edcc8581 | 815 | } |
1b0b46ee | 816 | process = &entry->subprocess.process; |
edcc8581 | 817 | |
42b0a86c | 818 | if (!(entry->supported_capabilities & wanted_capability)) |
edcc8581 LS |
819 | return 0; |
820 | ||
42b0a86c | 821 | if (wanted_capability & CAP_CLEAN) |
edcc8581 | 822 | filter_type = "clean"; |
42b0a86c | 823 | else if (wanted_capability & CAP_SMUDGE) |
edcc8581 LS |
824 | filter_type = "smudge"; |
825 | else | |
d26a328e | 826 | die(_("unexpected filter type")); |
edcc8581 LS |
827 | |
828 | sigchain_push(SIGPIPE, SIG_IGN); | |
829 | ||
830 | assert(strlen(filter_type) < LARGE_PACKET_DATA_MAX - strlen("command=\n")); | |
831 | err = packet_write_fmt_gently(process->in, "command=%s\n", filter_type); | |
832 | if (err) | |
833 | goto done; | |
834 | ||
835 | err = strlen(path) > LARGE_PACKET_DATA_MAX - strlen("pathname=\n"); | |
836 | if (err) { | |
d26a328e | 837 | error(_("path name too long for external filter")); |
edcc8581 LS |
838 | goto done; |
839 | } | |
840 | ||
841 | err = packet_write_fmt_gently(process->in, "pathname=%s\n", path); | |
842 | if (err) | |
843 | goto done; | |
844 | ||
ab90ecae | 845 | if (meta && meta->refname) { |
846 | err = packet_write_fmt_gently(process->in, "ref=%s\n", meta->refname); | |
847 | if (err) | |
848 | goto done; | |
849 | } | |
850 | ||
851 | if (meta && !is_null_oid(&meta->treeish)) { | |
852 | err = packet_write_fmt_gently(process->in, "treeish=%s\n", oid_to_hex(&meta->treeish)); | |
853 | if (err) | |
854 | goto done; | |
855 | } | |
856 | ||
857 | if (meta && !is_null_oid(&meta->blob)) { | |
858 | err = packet_write_fmt_gently(process->in, "blob=%s\n", oid_to_hex(&meta->blob)); | |
859 | if (err) | |
860 | goto done; | |
861 | } | |
862 | ||
2841e8f8 LS |
863 | if ((entry->supported_capabilities & CAP_DELAY) && |
864 | dco && dco->state == CE_CAN_DELAY) { | |
865 | can_delay = 1; | |
866 | err = packet_write_fmt_gently(process->in, "can-delay=1\n"); | |
867 | if (err) | |
868 | goto done; | |
869 | } | |
870 | ||
edcc8581 LS |
871 | err = packet_flush_gently(process->in); |
872 | if (err) | |
873 | goto done; | |
874 | ||
875 | if (fd >= 0) | |
3a63c6a4 | 876 | err = write_packetized_from_fd_no_flush(fd, process->in); |
edcc8581 | 877 | else |
3a63c6a4 JS |
878 | err = write_packetized_from_buf_no_flush(src, len, process->in); |
879 | if (err) | |
880 | goto done; | |
881 | ||
882 | err = packet_flush_gently(process->in); | |
edcc8581 LS |
883 | if (err) |
884 | goto done; | |
885 | ||
4f2a2e9f BP |
886 | err = subprocess_read_status(process->out, &filter_status); |
887 | if (err) | |
888 | goto done; | |
889 | ||
2841e8f8 LS |
890 | if (can_delay && !strcmp(filter_status.buf, "delayed")) { |
891 | string_list_insert(&dco->filters, cmd); | |
892 | string_list_insert(&dco->paths, path); | |
893 | } else { | |
894 | /* The filter got the blob and wants to send us a response. */ | |
895 | err = strcmp(filter_status.buf, "success"); | |
896 | if (err) | |
897 | goto done; | |
898 | ||
8c2efa5d JS |
899 | err = read_packetized_to_strbuf(process->out, &nbuf, |
900 | PACKET_READ_GENTLE_ON_EOF) < 0; | |
2841e8f8 LS |
901 | if (err) |
902 | goto done; | |
903 | ||
904 | err = subprocess_read_status(process->out, &filter_status); | |
905 | if (err) | |
906 | goto done; | |
907 | ||
908 | err = strcmp(filter_status.buf, "success"); | |
909 | } | |
910 | ||
911 | done: | |
912 | sigchain_pop(SIGPIPE); | |
913 | ||
914 | if (err) | |
915 | handle_filter_error(&filter_status, entry, wanted_capability); | |
916 | else | |
917 | strbuf_swap(dst, &nbuf); | |
918 | strbuf_release(&nbuf); | |
675ea4ef | 919 | strbuf_release(&filter_status); |
2841e8f8 LS |
920 | return !err; |
921 | } | |
922 | ||
923 | ||
924 | int async_query_available_blobs(const char *cmd, struct string_list *available_paths) | |
925 | { | |
926 | int err; | |
927 | char *line; | |
928 | struct cmd2process *entry; | |
929 | struct child_process *process; | |
930 | struct strbuf filter_status = STRBUF_INIT; | |
931 | ||
932 | assert(subprocess_map_initialized); | |
933 | entry = (struct cmd2process *)subprocess_find_entry(&subprocess_map, cmd); | |
934 | if (!entry) { | |
d26a328e NTND |
935 | error(_("external filter '%s' is not available anymore although " |
936 | "not all paths have been filtered"), cmd); | |
2841e8f8 LS |
937 | return 0; |
938 | } | |
939 | process = &entry->subprocess.process; | |
940 | sigchain_push(SIGPIPE, SIG_IGN); | |
941 | ||
942 | err = packet_write_fmt_gently( | |
943 | process->in, "command=list_available_blobs\n"); | |
edcc8581 LS |
944 | if (err) |
945 | goto done; | |
946 | ||
2841e8f8 | 947 | err = packet_flush_gently(process->in); |
edcc8581 LS |
948 | if (err) |
949 | goto done; | |
950 | ||
2841e8f8 LS |
951 | while ((line = packet_read_line(process->out, NULL))) { |
952 | const char *path; | |
953 | if (skip_prefix(line, "pathname=", &path)) | |
954 | string_list_insert(available_paths, xstrdup(path)); | |
955 | else | |
956 | ; /* ignore unknown keys */ | |
957 | } | |
958 | ||
4f2a2e9f BP |
959 | err = subprocess_read_status(process->out, &filter_status); |
960 | if (err) | |
961 | goto done; | |
962 | ||
edcc8581 LS |
963 | err = strcmp(filter_status.buf, "success"); |
964 | ||
965 | done: | |
966 | sigchain_pop(SIGPIPE); | |
967 | ||
9364fc29 | 968 | if (err) |
2841e8f8 | 969 | handle_filter_error(&filter_status, entry, 0); |
675ea4ef | 970 | strbuf_release(&filter_status); |
edcc8581 | 971 | return !err; |
aa4ed402 JH |
972 | } |
973 | ||
974 | static struct convert_driver { | |
975 | const char *name; | |
976 | struct convert_driver *next; | |
cd8be6c9 BH |
977 | const char *smudge; |
978 | const char *clean; | |
edcc8581 | 979 | const char *process; |
36daaaca | 980 | int required; |
aa4ed402 JH |
981 | } *user_convert, **user_convert_tail; |
982 | ||
234fa07e LS |
983 | static int apply_filter(const char *path, const char *src, size_t len, |
984 | int fd, struct strbuf *dst, struct convert_driver *drv, | |
2841e8f8 | 985 | const unsigned int wanted_capability, |
ab90ecae | 986 | const struct checkout_metadata *meta, |
2841e8f8 | 987 | struct delayed_checkout *dco) |
234fa07e LS |
988 | { |
989 | const char *cmd = NULL; | |
990 | ||
991 | if (!drv) | |
992 | return 0; | |
993 | ||
994 | if (!dst) | |
995 | return 1; | |
996 | ||
42b0a86c | 997 | if ((wanted_capability & CAP_CLEAN) && !drv->process && drv->clean) |
234fa07e | 998 | cmd = drv->clean; |
42b0a86c | 999 | else if ((wanted_capability & CAP_SMUDGE) && !drv->process && drv->smudge) |
234fa07e LS |
1000 | cmd = drv->smudge; |
1001 | ||
1002 | if (cmd && *cmd) | |
1003 | return apply_single_file_filter(path, src, len, fd, dst, cmd); | |
edcc8581 | 1004 | else if (drv->process && *drv->process) |
2841e8f8 | 1005 | return apply_multi_file_filter(path, src, len, fd, dst, |
ab90ecae | 1006 | drv->process, wanted_capability, meta, dco); |
234fa07e LS |
1007 | |
1008 | return 0; | |
1009 | } | |
1010 | ||
ef90d6d4 | 1011 | static int read_convert_config(const char *var, const char *value, void *cb) |
aa4ed402 | 1012 | { |
d731f0ad | 1013 | const char *key, *name; |
f5914f4b | 1014 | size_t namelen; |
aa4ed402 JH |
1015 | struct convert_driver *drv; |
1016 | ||
1017 | /* | |
1018 | * External conversion drivers are configured using | |
1019 | * "filter.<name>.variable". | |
1020 | */ | |
d731f0ad | 1021 | if (parse_config_key(var, "filter", &name, &namelen, &key) < 0 || !name) |
aa4ed402 | 1022 | return 0; |
aa4ed402 JH |
1023 | for (drv = user_convert; drv; drv = drv->next) |
1024 | if (!strncmp(drv->name, name, namelen) && !drv->name[namelen]) | |
1025 | break; | |
1026 | if (!drv) { | |
ca56dadb | 1027 | CALLOC_ARRAY(drv, 1); |
182af834 | 1028 | drv->name = xmemdupz(name, namelen); |
aa4ed402 JH |
1029 | *user_convert_tail = drv; |
1030 | user_convert_tail = &(drv->next); | |
1031 | } | |
1032 | ||
aa4ed402 JH |
1033 | /* |
1034 | * filter.<name>.smudge and filter.<name>.clean specifies | |
1035 | * the command line: | |
1036 | * | |
1037 | * command-line | |
1038 | * | |
1039 | * The command-line will not be interpolated in any way. | |
1040 | */ | |
1041 | ||
d731f0ad | 1042 | if (!strcmp("smudge", key)) |
cd8be6c9 BH |
1043 | return git_config_string(&drv->smudge, var, value); |
1044 | ||
d731f0ad | 1045 | if (!strcmp("clean", key)) |
cd8be6c9 | 1046 | return git_config_string(&drv->clean, var, value); |
aa4ed402 | 1047 | |
edcc8581 LS |
1048 | if (!strcmp("process", key)) |
1049 | return git_config_string(&drv->process, var, value); | |
1050 | ||
d731f0ad | 1051 | if (!strcmp("required", key)) { |
36daaaca JB |
1052 | drv->required = git_config_bool(var, value); |
1053 | return 0; | |
1054 | } | |
1055 | ||
aa4ed402 JH |
1056 | return 0; |
1057 | } | |
1058 | ||
3fed15f5 JH |
1059 | static int count_ident(const char *cp, unsigned long size) |
1060 | { | |
1061 | /* | |
af9b54bb | 1062 | * "$Id: 0000000000000000000000000000000000000000 $" <=> "$Id$" |
3fed15f5 JH |
1063 | */ |
1064 | int cnt = 0; | |
1065 | char ch; | |
1066 | ||
1067 | while (size) { | |
1068 | ch = *cp++; | |
1069 | size--; | |
1070 | if (ch != '$') | |
1071 | continue; | |
af9b54bb | 1072 | if (size < 3) |
3fed15f5 | 1073 | break; |
af9b54bb | 1074 | if (memcmp("Id", cp, 2)) |
3fed15f5 | 1075 | continue; |
af9b54bb AP |
1076 | ch = cp[2]; |
1077 | cp += 3; | |
1078 | size -= 3; | |
3fed15f5 | 1079 | if (ch == '$') |
af9b54bb | 1080 | cnt++; /* $Id$ */ |
3fed15f5 JH |
1081 | if (ch != ':') |
1082 | continue; | |
1083 | ||
1084 | /* | |
af9b54bb | 1085 | * "$Id: ... "; scan up to the closing dollar sign and discard. |
3fed15f5 JH |
1086 | */ |
1087 | while (size) { | |
1088 | ch = *cp++; | |
1089 | size--; | |
1090 | if (ch == '$') { | |
1091 | cnt++; | |
1092 | break; | |
1093 | } | |
a9f3049f HG |
1094 | if (ch == '\n') |
1095 | break; | |
3fed15f5 JH |
1096 | } |
1097 | } | |
1098 | return cnt; | |
1099 | } | |
1100 | ||
55ad152c | 1101 | static int ident_to_git(const char *src, size_t len, |
ec36c42a | 1102 | struct strbuf *buf, int ident) |
3fed15f5 | 1103 | { |
5ecd293d | 1104 | char *dst, *dollar; |
3fed15f5 | 1105 | |
4c3b57b9 | 1106 | if (!ident || (src && !count_ident(src, len))) |
5ecd293d PH |
1107 | return 0; |
1108 | ||
92ac3197 JK |
1109 | if (!buf) |
1110 | return 1; | |
1111 | ||
90d16ec0 PH |
1112 | /* only grow if not in place */ |
1113 | if (strbuf_avail(buf) + buf->len < len) | |
1114 | strbuf_grow(buf, len - buf->len); | |
5ecd293d PH |
1115 | dst = buf->buf; |
1116 | for (;;) { | |
1117 | dollar = memchr(src, '$', len); | |
1118 | if (!dollar) | |
1119 | break; | |
77321184 | 1120 | memmove(dst, src, dollar + 1 - src); |
5ecd293d PH |
1121 | dst += dollar + 1 - src; |
1122 | len -= dollar + 1 - src; | |
1123 | src = dollar + 1; | |
1124 | ||
1125 | if (len > 3 && !memcmp(src, "Id:", 3)) { | |
1126 | dollar = memchr(src + 3, '$', len - 3); | |
1127 | if (!dollar) | |
1128 | break; | |
a9f3049f HG |
1129 | if (memchr(src + 3, '\n', dollar - src - 3)) { |
1130 | /* Line break before the next dollar. */ | |
1131 | continue; | |
1132 | } | |
1133 | ||
af9b54bb AP |
1134 | memcpy(dst, "Id$", 3); |
1135 | dst += 3; | |
5ecd293d PH |
1136 | len -= dollar + 1 - src; |
1137 | src = dollar + 1; | |
3fed15f5 JH |
1138 | } |
1139 | } | |
77321184 | 1140 | memmove(dst, src, len); |
5ecd293d PH |
1141 | strbuf_setlen(buf, dst + len - buf->buf); |
1142 | return 1; | |
3fed15f5 JH |
1143 | } |
1144 | ||
55ad152c | 1145 | static int ident_to_worktree(const char *src, size_t len, |
ec36c42a | 1146 | struct strbuf *buf, int ident) |
3fed15f5 | 1147 | { |
f070facc | 1148 | struct object_id oid; |
07814d90 | 1149 | char *to_free = NULL, *dollar, *spc; |
5ecd293d | 1150 | int cnt; |
3fed15f5 JH |
1151 | |
1152 | if (!ident) | |
5ecd293d | 1153 | return 0; |
3fed15f5 | 1154 | |
5ecd293d | 1155 | cnt = count_ident(src, len); |
3fed15f5 | 1156 | if (!cnt) |
5ecd293d | 1157 | return 0; |
3fed15f5 | 1158 | |
5ecd293d PH |
1159 | /* are we "faking" in place editing ? */ |
1160 | if (src == buf->buf) | |
b315c5c0 | 1161 | to_free = strbuf_detach(buf, NULL); |
2dcde20e | 1162 | hash_object_file(the_hash_algo, src, len, "blob", &oid); |
3fed15f5 | 1163 | |
1a750441 | 1164 | strbuf_grow(buf, len + cnt * (the_hash_algo->hexsz + 3)); |
5ecd293d PH |
1165 | for (;;) { |
1166 | /* step 1: run to the next '$' */ | |
1167 | dollar = memchr(src, '$', len); | |
1168 | if (!dollar) | |
1169 | break; | |
1170 | strbuf_add(buf, src, dollar + 1 - src); | |
1171 | len -= dollar + 1 - src; | |
1172 | src = dollar + 1; | |
c23290d5 | 1173 | |
5ecd293d PH |
1174 | /* step 2: does it looks like a bit like Id:xxx$ or Id$ ? */ |
1175 | if (len < 3 || memcmp("Id", src, 2)) | |
3fed15f5 JH |
1176 | continue; |
1177 | ||
5ecd293d PH |
1178 | /* step 3: skip over Id$ or Id:xxxxx$ */ |
1179 | if (src[2] == '$') { | |
1180 | src += 3; | |
1181 | len -= 3; | |
1182 | } else if (src[2] == ':') { | |
1183 | /* | |
1184 | * It's possible that an expanded Id has crept its way into the | |
07814d90 HG |
1185 | * repository, we cope with that by stripping the expansion out. |
1186 | * This is probably not a good idea, since it will cause changes | |
1187 | * on checkout, which won't go away by stash, but let's keep it | |
1188 | * for git-style ids. | |
5ecd293d PH |
1189 | */ |
1190 | dollar = memchr(src + 3, '$', len - 3); | |
1191 | if (!dollar) { | |
1192 | /* incomplete keyword, no more '$', so just quit the loop */ | |
1193 | break; | |
1194 | } | |
c23290d5 | 1195 | |
a9f3049f HG |
1196 | if (memchr(src + 3, '\n', dollar - src - 3)) { |
1197 | /* Line break before the next dollar. */ | |
1198 | continue; | |
1199 | } | |
1200 | ||
07814d90 HG |
1201 | spc = memchr(src + 4, ' ', dollar - src - 4); |
1202 | if (spc && spc < dollar-1) { | |
1203 | /* There are spaces in unexpected places. | |
1204 | * This is probably an id from some other | |
1205 | * versioning system. Keep it for now. | |
1206 | */ | |
1207 | continue; | |
1208 | } | |
1209 | ||
5ecd293d PH |
1210 | len -= dollar + 1 - src; |
1211 | src = dollar + 1; | |
1212 | } else { | |
1213 | /* it wasn't a "Id$" or "Id:xxxx$" */ | |
1214 | continue; | |
1215 | } | |
c23290d5 | 1216 | |
5ecd293d PH |
1217 | /* step 4: substitute */ |
1218 | strbuf_addstr(buf, "Id: "); | |
f070facc | 1219 | strbuf_addstr(buf, oid_to_hex(&oid)); |
5ecd293d | 1220 | strbuf_addstr(buf, " $"); |
3fed15f5 | 1221 | } |
5ecd293d | 1222 | strbuf_add(buf, src, len); |
3fed15f5 | 1223 | |
5ecd293d PH |
1224 | free(to_free); |
1225 | return 1; | |
35ebfd6a JH |
1226 | } |
1227 | ||
107642fe LS |
1228 | static const char *git_path_check_encoding(struct attr_check_item *check) |
1229 | { | |
1230 | const char *value = check->value; | |
1231 | ||
1232 | if (ATTR_UNSET(value) || !strlen(value)) | |
1233 | return NULL; | |
1234 | ||
1235 | if (ATTR_TRUE(value) || ATTR_FALSE(value)) { | |
1236 | die(_("true/false are no valid working-tree-encodings")); | |
1237 | } | |
1238 | ||
1239 | /* Don't encode to the default encoding */ | |
1240 | if (same_encoding(value, default_encoding)) | |
1241 | return NULL; | |
1242 | ||
1243 | return value; | |
1244 | } | |
1245 | ||
38e95844 | 1246 | static enum convert_crlf_action git_path_check_crlf(struct attr_check_item *check) |
35ebfd6a | 1247 | { |
6073ee85 JH |
1248 | const char *value = check->value; |
1249 | ||
1250 | if (ATTR_TRUE(value)) | |
1251 | return CRLF_TEXT; | |
1252 | else if (ATTR_FALSE(value)) | |
1253 | return CRLF_BINARY; | |
1254 | else if (ATTR_UNSET(value)) | |
1255 | ; | |
1256 | else if (!strcmp(value, "input")) | |
df747b81 | 1257 | return CRLF_TEXT_INPUT; |
fd6cce9e EB |
1258 | else if (!strcmp(value, "auto")) |
1259 | return CRLF_AUTO; | |
df747b81 | 1260 | return CRLF_UNDEFINED; |
35ebfd6a JH |
1261 | } |
1262 | ||
7bd18054 | 1263 | static enum eol git_path_check_eol(struct attr_check_item *check) |
fd6cce9e EB |
1264 | { |
1265 | const char *value = check->value; | |
1266 | ||
1267 | if (ATTR_UNSET(value)) | |
1268 | ; | |
1269 | else if (!strcmp(value, "lf")) | |
1270 | return EOL_LF; | |
1271 | else if (!strcmp(value, "crlf")) | |
1272 | return EOL_CRLF; | |
1273 | return EOL_UNSET; | |
1274 | } | |
1275 | ||
7bd18054 | 1276 | static struct convert_driver *git_path_check_convert(struct attr_check_item *check) |
aa4ed402 JH |
1277 | { |
1278 | const char *value = check->value; | |
1279 | struct convert_driver *drv; | |
1280 | ||
1281 | if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value)) | |
1282 | return NULL; | |
1283 | for (drv = user_convert; drv; drv = drv->next) | |
1284 | if (!strcmp(value, drv->name)) | |
1285 | return drv; | |
1286 | return NULL; | |
1287 | } | |
1288 | ||
7bd18054 | 1289 | static int git_path_check_ident(struct attr_check_item *check) |
3fed15f5 JH |
1290 | { |
1291 | const char *value = check->value; | |
1292 | ||
1293 | return !!ATTR_TRUE(value); | |
1294 | } | |
1295 | ||
2c65d90f | 1296 | static struct attr_check *check; |
1297 | ||
8e978529 | 1298 | void convert_attrs(struct index_state *istate, |
38e95844 | 1299 | struct conv_attrs *ca, const char *path) |
83295964 | 1300 | { |
d64324cb | 1301 | struct attr_check_item *ccheck = NULL; |
83295964 | 1302 | |
2aef63d3 JH |
1303 | if (!check) { |
1304 | check = attr_check_initl("crlf", "ident", "filter", | |
107642fe LS |
1305 | "eol", "text", "working-tree-encoding", |
1306 | NULL); | |
83295964 JH |
1307 | user_convert_tail = &user_convert; |
1308 | git_config(read_convert_config, NULL); | |
1309 | } | |
3bfba20d | 1310 | |
d64324cb TB |
1311 | git_check_attr(istate, path, check); |
1312 | ccheck = check->items; | |
1313 | ca->crlf_action = git_path_check_crlf(ccheck + 4); | |
1314 | if (ca->crlf_action == CRLF_UNDEFINED) | |
1315 | ca->crlf_action = git_path_check_crlf(ccheck + 0); | |
1316 | ca->ident = git_path_check_ident(ccheck + 1); | |
1317 | ca->drv = git_path_check_convert(ccheck + 2); | |
1318 | if (ca->crlf_action != CRLF_BINARY) { | |
1319 | enum eol eol_attr = git_path_check_eol(ccheck + 3); | |
1320 | if (ca->crlf_action == CRLF_AUTO && eol_attr == EOL_LF) | |
1321 | ca->crlf_action = CRLF_AUTO_INPUT; | |
1322 | else if (ca->crlf_action == CRLF_AUTO && eol_attr == EOL_CRLF) | |
1323 | ca->crlf_action = CRLF_AUTO_CRLF; | |
1324 | else if (eol_attr == EOL_LF) | |
1325 | ca->crlf_action = CRLF_TEXT_INPUT; | |
1326 | else if (eol_attr == EOL_CRLF) | |
1327 | ca->crlf_action = CRLF_TEXT_CRLF; | |
3bfba20d | 1328 | } |
d64324cb | 1329 | ca->working_tree_encoding = git_path_check_encoding(ccheck + 5); |
5c94c93d MÅ |
1330 | |
1331 | /* Save attr and make a decision for action */ | |
1332 | ca->attr_action = ca->crlf_action; | |
df747b81 TB |
1333 | if (ca->crlf_action == CRLF_TEXT) |
1334 | ca->crlf_action = text_eol_is_crlf() ? CRLF_TEXT_CRLF : CRLF_TEXT_INPUT; | |
1335 | if (ca->crlf_action == CRLF_UNDEFINED && auto_crlf == AUTO_CRLF_FALSE) | |
1336 | ca->crlf_action = CRLF_BINARY; | |
1337 | if (ca->crlf_action == CRLF_UNDEFINED && auto_crlf == AUTO_CRLF_TRUE) | |
1338 | ca->crlf_action = CRLF_AUTO_CRLF; | |
1339 | if (ca->crlf_action == CRLF_UNDEFINED && auto_crlf == AUTO_CRLF_INPUT) | |
1340 | ca->crlf_action = CRLF_AUTO_INPUT; | |
83295964 JH |
1341 | } |
1342 | ||
2c65d90f | 1343 | void reset_parsed_attributes(void) |
1344 | { | |
1345 | struct convert_driver *drv, *next; | |
1346 | ||
1347 | attr_check_free(check); | |
1348 | check = NULL; | |
1349 | reset_merge_attributes(); | |
1350 | ||
1351 | for (drv = user_convert; drv; drv = next) { | |
1352 | next = drv->next; | |
1353 | free((void *)drv->name); | |
1354 | free(drv); | |
1355 | } | |
1356 | user_convert = NULL; | |
1357 | user_convert_tail = NULL; | |
1358 | } | |
1359 | ||
847a9e5d | 1360 | int would_convert_to_git_filter_fd(struct index_state *istate, const char *path) |
9035d75a SP |
1361 | { |
1362 | struct conv_attrs ca; | |
1363 | ||
7f944e26 | 1364 | convert_attrs(istate, &ca, path); |
9035d75a SP |
1365 | if (!ca.drv) |
1366 | return 0; | |
1367 | ||
1368 | /* | |
1369 | * Apply a filter to an fd only if the filter is required to succeed. | |
1370 | * We must die if the filter fails, because the original data before | |
1371 | * filtering is not available. | |
1372 | */ | |
1373 | if (!ca.drv->required) | |
1374 | return 0; | |
1375 | ||
ab90ecae | 1376 | return apply_filter(path, NULL, 0, -1, NULL, ca.drv, CAP_CLEAN, NULL, NULL); |
9035d75a SP |
1377 | } |
1378 | ||
847a9e5d | 1379 | const char *get_convert_attr_ascii(struct index_state *istate, const char *path) |
a7630bd4 TB |
1380 | { |
1381 | struct conv_attrs ca; | |
a7630bd4 | 1382 | |
7f944e26 | 1383 | convert_attrs(istate, &ca, path); |
bb211b4d | 1384 | switch (ca.attr_action) { |
df747b81 | 1385 | case CRLF_UNDEFINED: |
a7630bd4 TB |
1386 | return ""; |
1387 | case CRLF_BINARY: | |
1388 | return "-text"; | |
1389 | case CRLF_TEXT: | |
1390 | return "text"; | |
df747b81 | 1391 | case CRLF_TEXT_INPUT: |
a7630bd4 | 1392 | return "text eol=lf"; |
df747b81 TB |
1393 | case CRLF_TEXT_CRLF: |
1394 | return "text eol=crlf"; | |
a7630bd4 TB |
1395 | case CRLF_AUTO: |
1396 | return "text=auto"; | |
df747b81 | 1397 | case CRLF_AUTO_CRLF: |
65237284 | 1398 | return "text=auto eol=crlf"; |
df747b81 | 1399 | case CRLF_AUTO_INPUT: |
65237284 | 1400 | return "text=auto eol=lf"; |
a7630bd4 TB |
1401 | } |
1402 | return ""; | |
1403 | } | |
1404 | ||
847a9e5d | 1405 | int convert_to_git(struct index_state *istate, |
82b474e0 | 1406 | const char *path, const char *src, size_t len, |
8462ff43 | 1407 | struct strbuf *dst, int conv_flags) |
35ebfd6a | 1408 | { |
3bfba20d | 1409 | int ret = 0; |
3bfba20d | 1410 | struct conv_attrs ca; |
6073ee85 | 1411 | |
7f944e26 | 1412 | convert_attrs(istate, &ca, path); |
3fed15f5 | 1413 | |
ab90ecae | 1414 | ret |= apply_filter(path, src, len, -1, dst, ca.drv, CAP_CLEAN, NULL, NULL); |
234fa07e | 1415 | if (!ret && ca.drv && ca.drv->required) |
d26a328e | 1416 | die(_("%s: clean filter '%s' failed"), path, ca.drv->name); |
36daaaca | 1417 | |
92ac3197 | 1418 | if (ret && dst) { |
5ecd293d PH |
1419 | src = dst->buf; |
1420 | len = dst->len; | |
aa4ed402 | 1421 | } |
107642fe LS |
1422 | |
1423 | ret |= encode_to_git(path, src, len, dst, ca.working_tree_encoding, conv_flags); | |
1424 | if (ret && dst) { | |
1425 | src = dst->buf; | |
1426 | len = dst->len; | |
1427 | } | |
1428 | ||
8462ff43 TB |
1429 | if (!(conv_flags & CONV_EOL_KEEP_CRLF)) { |
1430 | ret |= crlf_to_git(istate, path, src, len, dst, ca.crlf_action, conv_flags); | |
2fea9de6 TB |
1431 | if (ret && dst) { |
1432 | src = dst->buf; | |
1433 | len = dst->len; | |
1434 | } | |
6073ee85 | 1435 | } |
55ad152c | 1436 | return ret | ident_to_git(src, len, dst, ca.ident); |
35ebfd6a JH |
1437 | } |
1438 | ||
847a9e5d | 1439 | void convert_to_git_filter_fd(struct index_state *istate, |
d6c41c20 | 1440 | const char *path, int fd, struct strbuf *dst, |
8462ff43 | 1441 | int conv_flags) |
9035d75a SP |
1442 | { |
1443 | struct conv_attrs ca; | |
7f944e26 | 1444 | convert_attrs(istate, &ca, path); |
9035d75a SP |
1445 | |
1446 | assert(ca.drv); | |
9035d75a | 1447 | |
ab90ecae | 1448 | if (!apply_filter(path, NULL, 0, fd, dst, ca.drv, CAP_CLEAN, NULL, NULL)) |
d26a328e | 1449 | die(_("%s: clean filter '%s' failed"), path, ca.drv->name); |
9035d75a | 1450 | |
107642fe | 1451 | encode_to_git(path, dst->buf, dst->len, dst, ca.working_tree_encoding, conv_flags); |
8462ff43 | 1452 | crlf_to_git(istate, path, dst->buf, dst->len, dst, ca.crlf_action, conv_flags); |
55ad152c | 1453 | ident_to_git(dst->buf, dst->len, dst, ca.ident); |
9035d75a SP |
1454 | } |
1455 | ||
55b4ad0e JH |
1456 | static int convert_to_working_tree_ca_internal(const struct conv_attrs *ca, |
1457 | const char *path, const char *src, | |
1458 | size_t len, struct strbuf *dst, | |
1459 | int normalizing, | |
1460 | const struct checkout_metadata *meta, | |
1461 | struct delayed_checkout *dco) | |
35ebfd6a | 1462 | { |
36daaaca | 1463 | int ret = 0, ret_filter = 0; |
3fed15f5 | 1464 | |
55b4ad0e | 1465 | ret |= ident_to_worktree(src, len, dst, ca->ident); |
5ecd293d PH |
1466 | if (ret) { |
1467 | src = dst->buf; | |
1468 | len = dst->len; | |
3fed15f5 | 1469 | } |
43dd2332 EB |
1470 | /* |
1471 | * CRLF conversion can be skipped if normalizing, unless there | |
edcc8581 LS |
1472 | * is a smudge or process filter (even if the process filter doesn't |
1473 | * support smudge). The filters might expect CRLFs. | |
43dd2332 | 1474 | */ |
55b4ad0e JH |
1475 | if ((ca->drv && (ca->drv->smudge || ca->drv->process)) || !normalizing) { |
1476 | ret |= crlf_to_worktree(src, len, dst, ca->crlf_action); | |
43dd2332 EB |
1477 | if (ret) { |
1478 | src = dst->buf; | |
1479 | len = dst->len; | |
1480 | } | |
aa4ed402 | 1481 | } |
36daaaca | 1482 | |
55b4ad0e | 1483 | ret |= encode_to_worktree(path, src, len, dst, ca->working_tree_encoding); |
107642fe LS |
1484 | if (ret) { |
1485 | src = dst->buf; | |
1486 | len = dst->len; | |
1487 | } | |
1488 | ||
2841e8f8 | 1489 | ret_filter = apply_filter( |
55b4ad0e JH |
1490 | path, src, len, -1, dst, ca->drv, CAP_SMUDGE, meta, dco); |
1491 | if (!ret_filter && ca->drv && ca->drv->required) | |
1492 | die(_("%s: smudge filter %s failed"), path, ca->drv->name); | |
36daaaca JB |
1493 | |
1494 | return ret | ret_filter; | |
35ebfd6a | 1495 | } |
f217f0e8 | 1496 | |
55b4ad0e JH |
1497 | int async_convert_to_working_tree_ca(const struct conv_attrs *ca, |
1498 | const char *path, const char *src, | |
1499 | size_t len, struct strbuf *dst, | |
1500 | const struct checkout_metadata *meta, | |
1501 | void *dco) | |
2841e8f8 | 1502 | { |
55b4ad0e JH |
1503 | return convert_to_working_tree_ca_internal(ca, path, src, len, dst, 0, |
1504 | meta, dco); | |
2841e8f8 LS |
1505 | } |
1506 | ||
55b4ad0e JH |
1507 | int convert_to_working_tree_ca(const struct conv_attrs *ca, |
1508 | const char *path, const char *src, | |
1509 | size_t len, struct strbuf *dst, | |
1510 | const struct checkout_metadata *meta) | |
43dd2332 | 1511 | { |
55b4ad0e JH |
1512 | return convert_to_working_tree_ca_internal(ca, path, src, len, dst, 0, |
1513 | meta, NULL); | |
43dd2332 EB |
1514 | } |
1515 | ||
847a9e5d | 1516 | int renormalize_buffer(struct index_state *istate, const char *path, |
a33e0b2a | 1517 | const char *src, size_t len, struct strbuf *dst) |
f217f0e8 | 1518 | { |
55b4ad0e JH |
1519 | struct conv_attrs ca; |
1520 | int ret; | |
1521 | ||
1522 | convert_attrs(istate, &ca, path); | |
1523 | ret = convert_to_working_tree_ca_internal(&ca, path, src, len, dst, 1, | |
1524 | NULL, NULL); | |
f217f0e8 EB |
1525 | if (ret) { |
1526 | src = dst->buf; | |
1527 | len = dst->len; | |
1528 | } | |
8462ff43 | 1529 | return ret | convert_to_git(istate, path, src, len, dst, CONV_EOL_RENORMALIZE); |
f217f0e8 | 1530 | } |
dd8e9121 | 1531 | |
b6691092 JH |
1532 | /***************************************************************** |
1533 | * | |
749f763d | 1534 | * Streaming conversion support |
b6691092 JH |
1535 | * |
1536 | *****************************************************************/ | |
1537 | ||
1538 | typedef int (*filter_fn)(struct stream_filter *, | |
1539 | const char *input, size_t *isize_p, | |
1540 | char *output, size_t *osize_p); | |
1541 | typedef void (*free_fn)(struct stream_filter *); | |
1542 | ||
1543 | struct stream_filter_vtbl { | |
1544 | filter_fn filter; | |
1545 | free_fn free; | |
1546 | }; | |
1547 | ||
1548 | struct stream_filter { | |
1549 | struct stream_filter_vtbl *vtbl; | |
1550 | }; | |
1551 | ||
1552 | static int null_filter_fn(struct stream_filter *filter, | |
1553 | const char *input, size_t *isize_p, | |
1554 | char *output, size_t *osize_p) | |
1555 | { | |
4ae66704 JH |
1556 | size_t count; |
1557 | ||
1558 | if (!input) | |
1559 | return 0; /* we do not keep any states */ | |
1560 | count = *isize_p; | |
b6691092 JH |
1561 | if (*osize_p < count) |
1562 | count = *osize_p; | |
1563 | if (count) { | |
1564 | memmove(output, input, count); | |
1565 | *isize_p -= count; | |
1566 | *osize_p -= count; | |
1567 | } | |
1568 | return 0; | |
1569 | } | |
1570 | ||
1571 | static void null_free_fn(struct stream_filter *filter) | |
1572 | { | |
1573 | ; /* nothing -- null instances are shared */ | |
1574 | } | |
1575 | ||
1576 | static struct stream_filter_vtbl null_vtbl = { | |
1577 | null_filter_fn, | |
1578 | null_free_fn, | |
1579 | }; | |
1580 | ||
1581 | static struct stream_filter null_filter_singleton = { | |
1582 | &null_vtbl, | |
1583 | }; | |
1584 | ||
1585 | int is_null_stream_filter(struct stream_filter *filter) | |
1586 | { | |
1587 | return filter == &null_filter_singleton; | |
1588 | } | |
1589 | ||
b84c7839 JH |
1590 | |
1591 | /* | |
1592 | * LF-to-CRLF filter | |
1593 | */ | |
284e3d28 CMN |
1594 | |
1595 | struct lf_to_crlf_filter { | |
1596 | struct stream_filter filter; | |
8496f568 JH |
1597 | unsigned has_held:1; |
1598 | char held; | |
284e3d28 CMN |
1599 | }; |
1600 | ||
e322ee38 JH |
1601 | static int lf_to_crlf_filter_fn(struct stream_filter *filter, |
1602 | const char *input, size_t *isize_p, | |
1603 | char *output, size_t *osize_p) | |
1604 | { | |
284e3d28 CMN |
1605 | size_t count, o = 0; |
1606 | struct lf_to_crlf_filter *lf_to_crlf = (struct lf_to_crlf_filter *)filter; | |
1607 | ||
8496f568 JH |
1608 | /* |
1609 | * We may be holding onto the CR to see if it is followed by a | |
1610 | * LF, in which case we would need to go to the main loop. | |
1611 | * Otherwise, just emit it to the output stream. | |
1612 | */ | |
1613 | if (lf_to_crlf->has_held && (lf_to_crlf->held != '\r' || !input)) { | |
1614 | output[o++] = lf_to_crlf->held; | |
1615 | lf_to_crlf->has_held = 0; | |
284e3d28 | 1616 | } |
e322ee38 | 1617 | |
87afe9a5 JH |
1618 | /* We are told to drain */ |
1619 | if (!input) { | |
1620 | *osize_p -= o; | |
1621 | return 0; | |
1622 | } | |
e322ee38 | 1623 | |
e322ee38 | 1624 | count = *isize_p; |
8496f568 | 1625 | if (count || lf_to_crlf->has_held) { |
284e3d28 | 1626 | size_t i; |
8496f568 JH |
1627 | int was_cr = 0; |
1628 | ||
1629 | if (lf_to_crlf->has_held) { | |
1630 | was_cr = 1; | |
1631 | lf_to_crlf->has_held = 0; | |
1632 | } | |
1633 | ||
284e3d28 | 1634 | for (i = 0; o < *osize_p && i < count; i++) { |
e322ee38 | 1635 | char ch = input[i]; |
8496f568 | 1636 | |
e322ee38 | 1637 | if (ch == '\n') { |
284e3d28 | 1638 | output[o++] = '\r'; |
8496f568 JH |
1639 | } else if (was_cr) { |
1640 | /* | |
1641 | * Previous round saw CR and it is not followed | |
1642 | * by a LF; emit the CR before processing the | |
1643 | * current character. | |
1644 | */ | |
1645 | output[o++] = '\r'; | |
e322ee38 | 1646 | } |
8496f568 JH |
1647 | |
1648 | /* | |
1649 | * We may have consumed the last output slot, | |
1650 | * in which case we need to break out of this | |
1651 | * loop; hold the current character before | |
1652 | * returning. | |
1653 | */ | |
1654 | if (*osize_p <= o) { | |
1655 | lf_to_crlf->has_held = 1; | |
1656 | lf_to_crlf->held = ch; | |
1657 | continue; /* break but increment i */ | |
1658 | } | |
1659 | ||
1660 | if (ch == '\r') { | |
1661 | was_cr = 1; | |
1662 | continue; | |
1663 | } | |
1664 | ||
1665 | was_cr = 0; | |
e322ee38 JH |
1666 | output[o++] = ch; |
1667 | } | |
1668 | ||
1669 | *osize_p -= o; | |
1670 | *isize_p -= i; | |
8496f568 JH |
1671 | |
1672 | if (!lf_to_crlf->has_held && was_cr) { | |
1673 | lf_to_crlf->has_held = 1; | |
1674 | lf_to_crlf->held = '\r'; | |
1675 | } | |
e322ee38 JH |
1676 | } |
1677 | return 0; | |
1678 | } | |
1679 | ||
284e3d28 CMN |
1680 | static void lf_to_crlf_free_fn(struct stream_filter *filter) |
1681 | { | |
1682 | free(filter); | |
1683 | } | |
1684 | ||
e322ee38 JH |
1685 | static struct stream_filter_vtbl lf_to_crlf_vtbl = { |
1686 | lf_to_crlf_filter_fn, | |
284e3d28 | 1687 | lf_to_crlf_free_fn, |
e322ee38 JH |
1688 | }; |
1689 | ||
284e3d28 CMN |
1690 | static struct stream_filter *lf_to_crlf_filter(void) |
1691 | { | |
87afe9a5 | 1692 | struct lf_to_crlf_filter *lf_to_crlf = xcalloc(1, sizeof(*lf_to_crlf)); |
e322ee38 | 1693 | |
284e3d28 | 1694 | lf_to_crlf->filter.vtbl = &lf_to_crlf_vtbl; |
284e3d28 CMN |
1695 | return (struct stream_filter *)lf_to_crlf; |
1696 | } | |
b84c7839 | 1697 | |
a265a7f9 JH |
1698 | /* |
1699 | * Cascade filter | |
1700 | */ | |
1701 | #define FILTER_BUFFER 1024 | |
1702 | struct cascade_filter { | |
1703 | struct stream_filter filter; | |
1704 | struct stream_filter *one; | |
1705 | struct stream_filter *two; | |
1706 | char buf[FILTER_BUFFER]; | |
1707 | int end, ptr; | |
1708 | }; | |
1709 | ||
1710 | static int cascade_filter_fn(struct stream_filter *filter, | |
1711 | const char *input, size_t *isize_p, | |
1712 | char *output, size_t *osize_p) | |
1713 | { | |
1714 | struct cascade_filter *cas = (struct cascade_filter *) filter; | |
1715 | size_t filled = 0; | |
1716 | size_t sz = *osize_p; | |
1717 | size_t to_feed, remaining; | |
1718 | ||
1719 | /* | |
1720 | * input -- (one) --> buf -- (two) --> output | |
1721 | */ | |
1722 | while (filled < sz) { | |
1723 | remaining = sz - filled; | |
1724 | ||
1725 | /* do we already have something to feed two with? */ | |
1726 | if (cas->ptr < cas->end) { | |
1727 | to_feed = cas->end - cas->ptr; | |
1728 | if (stream_filter(cas->two, | |
1729 | cas->buf + cas->ptr, &to_feed, | |
1730 | output + filled, &remaining)) | |
1731 | return -1; | |
1732 | cas->ptr += (cas->end - cas->ptr) - to_feed; | |
1733 | filled = sz - remaining; | |
1734 | continue; | |
1735 | } | |
1736 | ||
1737 | /* feed one from upstream and have it emit into our buffer */ | |
1738 | to_feed = input ? *isize_p : 0; | |
1739 | if (input && !to_feed) | |
1740 | break; | |
1741 | remaining = sizeof(cas->buf); | |
1742 | if (stream_filter(cas->one, | |
1743 | input, &to_feed, | |
1744 | cas->buf, &remaining)) | |
1745 | return -1; | |
1746 | cas->end = sizeof(cas->buf) - remaining; | |
1747 | cas->ptr = 0; | |
1748 | if (input) { | |
1749 | size_t fed = *isize_p - to_feed; | |
1750 | *isize_p -= fed; | |
1751 | input += fed; | |
1752 | } | |
1753 | ||
1754 | /* do we know that we drained one completely? */ | |
1755 | if (input || cas->end) | |
1756 | continue; | |
1757 | ||
1758 | /* tell two to drain; we have nothing more to give it */ | |
1759 | to_feed = 0; | |
1760 | remaining = sz - filled; | |
1761 | if (stream_filter(cas->two, | |
1762 | NULL, &to_feed, | |
1763 | output + filled, &remaining)) | |
1764 | return -1; | |
1765 | if (remaining == (sz - filled)) | |
1766 | break; /* completely drained two */ | |
1767 | filled = sz - remaining; | |
1768 | } | |
1769 | *osize_p -= filled; | |
1770 | return 0; | |
1771 | } | |
1772 | ||
1773 | static void cascade_free_fn(struct stream_filter *filter) | |
1774 | { | |
1775 | struct cascade_filter *cas = (struct cascade_filter *)filter; | |
1776 | free_stream_filter(cas->one); | |
1777 | free_stream_filter(cas->two); | |
1778 | free(filter); | |
1779 | } | |
1780 | ||
1781 | static struct stream_filter_vtbl cascade_vtbl = { | |
1782 | cascade_filter_fn, | |
1783 | cascade_free_fn, | |
1784 | }; | |
1785 | ||
1786 | static struct stream_filter *cascade_filter(struct stream_filter *one, | |
1787 | struct stream_filter *two) | |
1788 | { | |
1789 | struct cascade_filter *cascade; | |
1790 | ||
1791 | if (!one || is_null_stream_filter(one)) | |
1792 | return two; | |
1793 | if (!two || is_null_stream_filter(two)) | |
1794 | return one; | |
1795 | ||
1796 | cascade = xmalloc(sizeof(*cascade)); | |
1797 | cascade->one = one; | |
1798 | cascade->two = two; | |
1799 | cascade->end = cascade->ptr = 0; | |
1800 | cascade->filter.vtbl = &cascade_vtbl; | |
1801 | return (struct stream_filter *)cascade; | |
1802 | } | |
1803 | ||
b84c7839 JH |
1804 | /* |
1805 | * ident filter | |
1806 | */ | |
1807 | #define IDENT_DRAINING (-1) | |
1808 | #define IDENT_SKIPPING (-2) | |
1809 | struct ident_filter { | |
1810 | struct stream_filter filter; | |
1811 | struct strbuf left; | |
1812 | int state; | |
1a750441 | 1813 | char ident[GIT_MAX_HEXSZ + 5]; /* ": x40 $" */ |
b84c7839 JH |
1814 | }; |
1815 | ||
1816 | static int is_foreign_ident(const char *str) | |
1817 | { | |
1818 | int i; | |
1819 | ||
ae021d87 | 1820 | if (!skip_prefix(str, "$Id: ", &str)) |
b84c7839 | 1821 | return 0; |
ae021d87 | 1822 | for (i = 0; str[i]; i++) { |
b84c7839 JH |
1823 | if (isspace(str[i]) && str[i+1] != '$') |
1824 | return 1; | |
1825 | } | |
1826 | return 0; | |
1827 | } | |
1828 | ||
1829 | static void ident_drain(struct ident_filter *ident, char **output_p, size_t *osize_p) | |
1830 | { | |
1831 | size_t to_drain = ident->left.len; | |
1832 | ||
1833 | if (*osize_p < to_drain) | |
1834 | to_drain = *osize_p; | |
1835 | if (to_drain) { | |
1836 | memcpy(*output_p, ident->left.buf, to_drain); | |
1837 | strbuf_remove(&ident->left, 0, to_drain); | |
1838 | *output_p += to_drain; | |
1839 | *osize_p -= to_drain; | |
1840 | } | |
1841 | if (!ident->left.len) | |
1842 | ident->state = 0; | |
1843 | } | |
1844 | ||
1845 | static int ident_filter_fn(struct stream_filter *filter, | |
1846 | const char *input, size_t *isize_p, | |
1847 | char *output, size_t *osize_p) | |
1848 | { | |
1849 | struct ident_filter *ident = (struct ident_filter *)filter; | |
1850 | static const char head[] = "$Id"; | |
1851 | ||
1852 | if (!input) { | |
1853 | /* drain upon eof */ | |
1854 | switch (ident->state) { | |
1855 | default: | |
1856 | strbuf_add(&ident->left, head, ident->state); | |
1cf01a34 | 1857 | /* fallthrough */ |
b84c7839 | 1858 | case IDENT_SKIPPING: |
1cf01a34 | 1859 | /* fallthrough */ |
b84c7839 JH |
1860 | case IDENT_DRAINING: |
1861 | ident_drain(ident, &output, osize_p); | |
1862 | } | |
1863 | return 0; | |
1864 | } | |
1865 | ||
1866 | while (*isize_p || (ident->state == IDENT_DRAINING)) { | |
1867 | int ch; | |
1868 | ||
1869 | if (ident->state == IDENT_DRAINING) { | |
1870 | ident_drain(ident, &output, osize_p); | |
1871 | if (!*osize_p) | |
1872 | break; | |
1873 | continue; | |
1874 | } | |
1875 | ||
1876 | ch = *(input++); | |
1877 | (*isize_p)--; | |
1878 | ||
1879 | if (ident->state == IDENT_SKIPPING) { | |
1880 | /* | |
1881 | * Skipping until '$' or LF, but keeping them | |
1882 | * in case it is a foreign ident. | |
1883 | */ | |
1884 | strbuf_addch(&ident->left, ch); | |
1885 | if (ch != '\n' && ch != '$') | |
1886 | continue; | |
1887 | if (ch == '$' && !is_foreign_ident(ident->left.buf)) { | |
1888 | strbuf_setlen(&ident->left, sizeof(head) - 1); | |
1889 | strbuf_addstr(&ident->left, ident->ident); | |
1890 | } | |
1891 | ident->state = IDENT_DRAINING; | |
1892 | continue; | |
1893 | } | |
1894 | ||
1895 | if (ident->state < sizeof(head) && | |
1896 | head[ident->state] == ch) { | |
1897 | ident->state++; | |
1898 | continue; | |
1899 | } | |
1900 | ||
1901 | if (ident->state) | |
1902 | strbuf_add(&ident->left, head, ident->state); | |
1903 | if (ident->state == sizeof(head) - 1) { | |
1904 | if (ch != ':' && ch != '$') { | |
1905 | strbuf_addch(&ident->left, ch); | |
1906 | ident->state = 0; | |
1907 | continue; | |
1908 | } | |
1909 | ||
1910 | if (ch == ':') { | |
1911 | strbuf_addch(&ident->left, ch); | |
1912 | ident->state = IDENT_SKIPPING; | |
1913 | } else { | |
1914 | strbuf_addstr(&ident->left, ident->ident); | |
1915 | ident->state = IDENT_DRAINING; | |
1916 | } | |
1917 | continue; | |
1918 | } | |
1919 | ||
1920 | strbuf_addch(&ident->left, ch); | |
1921 | ident->state = IDENT_DRAINING; | |
1922 | } | |
1923 | return 0; | |
1924 | } | |
1925 | ||
1926 | static void ident_free_fn(struct stream_filter *filter) | |
1927 | { | |
1928 | struct ident_filter *ident = (struct ident_filter *)filter; | |
1929 | strbuf_release(&ident->left); | |
1930 | free(filter); | |
1931 | } | |
1932 | ||
1933 | static struct stream_filter_vtbl ident_vtbl = { | |
1934 | ident_filter_fn, | |
1935 | ident_free_fn, | |
1936 | }; | |
1937 | ||
1a750441 | 1938 | static struct stream_filter *ident_filter(const struct object_id *oid) |
b84c7839 JH |
1939 | { |
1940 | struct ident_filter *ident = xmalloc(sizeof(*ident)); | |
1941 | ||
5096d490 | 1942 | xsnprintf(ident->ident, sizeof(ident->ident), |
1a750441 | 1943 | ": %s $", oid_to_hex(oid)); |
b84c7839 JH |
1944 | strbuf_init(&ident->left, 0); |
1945 | ident->filter.vtbl = &ident_vtbl; | |
1946 | ident->state = 0; | |
1947 | return (struct stream_filter *)ident; | |
1948 | } | |
1949 | ||
dd8e9121 | 1950 | /* |
3e9e82c0 | 1951 | * Return an appropriately constructed filter for the given ca, or NULL if |
b6691092 JH |
1952 | * the contents cannot be filtered without reading the whole thing |
1953 | * in-core. | |
1954 | * | |
2b0f19fa | 1955 | * Note that you would be crazy to set CRLF, smudge/clean or ident to a |
b6691092 | 1956 | * large binary blob you would want us not to slurp into the memory! |
dd8e9121 | 1957 | */ |
3e9e82c0 JH |
1958 | struct stream_filter *get_stream_filter_ca(const struct conv_attrs *ca, |
1959 | const struct object_id *oid) | |
dd8e9121 | 1960 | { |
b84c7839 | 1961 | struct stream_filter *filter = NULL; |
dd8e9121 | 1962 | |
f59d15bb | 1963 | if (classify_conv_attrs(ca) != CA_CLASS_STREAMABLE) |
107642fe LS |
1964 | return NULL; |
1965 | ||
3e9e82c0 | 1966 | if (ca->ident) |
1a750441 | 1967 | filter = ident_filter(oid); |
dd8e9121 | 1968 | |
3e9e82c0 | 1969 | if (output_eol(ca->crlf_action) == EOL_CRLF) |
284e3d28 | 1970 | filter = cascade_filter(filter, lf_to_crlf_filter()); |
caa47adc TB |
1971 | else |
1972 | filter = cascade_filter(filter, &null_filter_singleton); | |
e322ee38 | 1973 | |
b84c7839 | 1974 | return filter; |
b6691092 JH |
1975 | } |
1976 | ||
8e978529 | 1977 | struct stream_filter *get_stream_filter(struct index_state *istate, |
3e9e82c0 JH |
1978 | const char *path, |
1979 | const struct object_id *oid) | |
1980 | { | |
1981 | struct conv_attrs ca; | |
1982 | convert_attrs(istate, &ca, path); | |
1983 | return get_stream_filter_ca(&ca, oid); | |
1984 | } | |
1985 | ||
b6691092 JH |
1986 | void free_stream_filter(struct stream_filter *filter) |
1987 | { | |
1988 | filter->vtbl->free(filter); | |
1989 | } | |
1990 | ||
1991 | int stream_filter(struct stream_filter *filter, | |
1992 | const char *input, size_t *isize_p, | |
1993 | char *output, size_t *osize_p) | |
1994 | { | |
1995 | return filter->vtbl->filter(filter, input, isize_p, output, osize_p); | |
dd8e9121 | 1996 | } |
c397aac0 | 1997 | |
1998 | void init_checkout_metadata(struct checkout_metadata *meta, const char *refname, | |
1999 | const struct object_id *treeish, | |
2000 | const struct object_id *blob) | |
2001 | { | |
2002 | memset(meta, 0, sizeof(*meta)); | |
2003 | if (refname) | |
2004 | meta->refname = refname; | |
2005 | if (treeish) | |
2006 | oidcpy(&meta->treeish, treeish); | |
2007 | if (blob) | |
2008 | oidcpy(&meta->blob, blob); | |
2009 | } | |
2010 | ||
2011 | void clone_checkout_metadata(struct checkout_metadata *dst, | |
2012 | const struct checkout_metadata *src, | |
2013 | const struct object_id *blob) | |
2014 | { | |
2015 | memcpy(dst, src, sizeof(*dst)); | |
2016 | if (blob) | |
2017 | oidcpy(&dst->blob, blob); | |
2018 | } | |
f59d15bb JH |
2019 | |
2020 | enum conv_attrs_classification classify_conv_attrs(const struct conv_attrs *ca) | |
2021 | { | |
2022 | if (ca->drv) { | |
2023 | if (ca->drv->process) | |
2024 | return CA_CLASS_INCORE_PROCESS; | |
2025 | if (ca->drv->smudge || ca->drv->clean) | |
2026 | return CA_CLASS_INCORE_FILTER; | |
2027 | } | |
2028 | ||
2029 | if (ca->working_tree_encoding) | |
2030 | return CA_CLASS_INCORE; | |
2031 | ||
2032 | if (ca->crlf_action == CRLF_AUTO || ca->crlf_action == CRLF_AUTO_CRLF) | |
2033 | return CA_CLASS_INCORE; | |
2034 | ||
2035 | return CA_CLASS_STREAMABLE; | |
2036 | } |